Klebsiella pneumoniae is an important opportunistic pathogen and a major public health concern. K. pneumoniae is known to disseminate within the hospital environment and cause opportunistic infections in vulnerable patients, but may also cause severe community-acquired infections such as liver abscesses and pneumonia. K. pneumoniae is separated into seven subspecies, all part of the K. pneumoniae species complex (KpSC), where K. pneumoniae subspecies pneumoniae has been found to have the highest public health burden (Wyres et al., 2020). Several clones of K. pneumoniae have been defined as global problem clones (Wyres et al., 2020). These are generally separated into two groups, hypervirulent- and multidrug-resistant clones, with little overlap between the two groups. However, convergence of virulence and multidrug-resistance has been described and is of major public health concern (Lam et al., 2019; Lan et al., 2021).

Virulence in K. pneumoniae is often associated with the presence of the siderophores aerobactin, salmochelin, and/or yersiniabactin, in addition to the genotoxin colibactin and the hypermucoviscosity-encoding locus rmpADC. Aerobactin, salmochelin, yersiniabactin, and colibactin are siderophore systems that produce iron-chelating molecules, enabling iron-scavenging from various sources (Nassif and Sansonetti, 1986; Carniel, 2001; Müller et al., 2009; Putze et al., 2009). Aerobactin has been identified as the major contributor to virulence (Matono et al., 2022), even in the presence of additional virulence factors (Russo et al., 2014, 2015). The aerobactin, salmochelin, and yersiniabactin loci are classified into several subtypes (Lam et al., 2021), making it possible to track different lineages of each locus. The aerobactin and salmochelin loci are mobilized by both conserved and diverse plasmid structures (Lam et al., 2018b), while yersiniabactin is mobilized by an integrative conjugative element often found on the chromosome (Lam et al., 2018a).

Klebsiella pneumoniae is well studied in relation to human infection. However, there is a lack of knowledge regarding the presence and pathogenicity of K. pneumoniae in other niches. Gaining insight into other niches that might harbor K. pneumoniae is of public health interest and helps us to understand the dynamics of this bacterium in a One Health perspective. In 2018, we isolated K. pneumoniae from cecal samples from healthy broiler and turkey flocks in Norway, and characterized the isolates with regard to population structure, antimicrobial resistance, and virulence genes (Franklin-Alming et al., 2021). A higher occurrence of K. pneumoniae was detected in the turkey flocks compared to broiler flocks, and specific sequence types (STs) were predominant among the isolates, including ST35 and ST37. Whether these STs persisted and dominated over time could not be determined in the 2018 study as sampling was performed within a single year.

In the 2018 study, a clonal expansion of a hypervirulent ST4710 strain was observed in the turkey samples. This clone carried yersiniabactin (ybt16) on the chromosome, in addition to an IncFII plasmid, p4710, harboring both aerobactin (iuc5) and salmochelin (iro5), contributing to a hypervirulent phenotype. K. pneumoniae is occasionally associated with infections in animals, including poultry. An ST4710 isolate was also detected from a clinical air sac sample submitted to our diagnostic laboratory in 2018, from a turkey in a flock with several cases of airsacculitis. Microbiological analysis revealed rich growth of mixed flora with Escherichia coli, K. pneumoniae and Candida krusei, indicating an intestinal origin of the infection.

The aim of the present study was to explore potential shifts in the population dynamics, virulence, and antimicrobial resistance of K. pneumoniae isolates from broiler and turkey flocks over time by comparing isolates from 2018 to 2020. Lastly, we wanted to characterize the clinical ST4710 isolate and compare it to intestinal ST4710 isolates from healthy birds.

This study aimed to evaluate K. pneumoniae ST dynamics and presence of AMR- and virulence determinants in broiler and turkey flocks over time by comparing genomes from 2018 to 2020. Overall, we identified a high diversity of K. pneumoniae among broiler- and turkey flocks in Norway both in 2018 (Franklin-Alming et al., 2021) and 2020. Several STs were detected both in 2018 and 2020, some of which were more diverse in broilers than in turkeys. Further, we discovered that the ST4710 hypervirulent clone observed in turkeys in 2018 was also present, but less prevalent in turkeys in 2020. However, the ST4710 isolates from 2020 lacked the salmochelin operon. Lastly, we characterized a clinical air sac ST4710 isolate, and found that it was closely related (<12 SNPs) to the intestinal ST4710 isolates from 2018.

The major STs previously identified in 2018 were still dominating in 2020. This indicates persistence of these major STs within the poultry production. This may be due to local re-circulation or introduction through the poultry production chain from parent flocks. The overall within-ST diversity for these major STs was higher in broilers compared to turkeys. This was also shown for the ST37 clade. These findings were supported by the result from the core genome analysis, which included a much bigger portion of each genome, and the inclusion of the locus variant STs within the same clade. The clustering by host species identified in ST37 suggest little to no overlap of this ST between these two animal niches. For ST37, the smallest within-host SNP distances were detected in the turkey samples, where both sampling years intermingled within the same clade. This suggests that ST37 is persisting in the turkey population within the study period. The high similarity between the turkey ST37 isolates can likely be attributed to clonal expansion. This expansion may be due to reintroduction from higher up in the poultry production pyramid, which has been previously described for antimicrobial resistant E. coli in the Norwegian poultry production (Mo et al., 2014; Kaspersen et al., 2020a,b).

The lower diversity seen in ST37 from turkeys may also be explained by the population and production differences between turkeys and broilers. The turkey production in Norway is small with only about 50 farms all together producing about 950,000 turkeys, while the broiler production is much larger with about 500 farms producing more than 74 million broilers (Falk et al., 2022). The sampling strategy used in NORM-VET was the same for the 2 years, therefore the results from the 2 years should be comparable. However, the population and production differences may have an impact on the detected diversity between the two species in the study. With a sampling design aiming at sampling 300 random broiler flocks per year, there will be very few broiler farms from which samples are taken from more than one flock. As a contrast to samples from turkey, where almost all produced turkey flocks are sampled per year, and therefore consist of samples from the same farm though from several flocks. In addition, due to the relatively small turkey population, several of the turkey flocks probably share the same parent animal flock, while this is less of a concern in broilers. There is also a difference with regard to time of slaughter and thereby age of sampling between the two populations. Broilers are slaughtered at approximately 30 days of age, while turkeys are slaughtered at 87–130 days (Animalia, 2019). All these aspects may have had an impact on the results, and thereby explain the difference in occurrence of K. pneumoniae between the two species.

Several STs from Norwegian poultry overlap with STs isolated from both healthy and sick humans in Norway, such as ST35, ST37, and ST17 (Raffelsberger et al., 2021; Fostervold et al., 2022; Hetland et al., 2023a). AMR- and virulence genes were scarce within each of these STs in the current study. However, some of the ST35 isolates from both human clinical samples and carriage studies harbored yersiniabactin ybt8, similar to the two yersiniabactin-positive ST35-isolates from the current study, indicating that there might be a potential for overlap between the human and poultry niches. This is currently being followed up in an ongoing study investigating the cross-talk of K. pneumoniae between several niches and host species, such as humans, animals, and the environment (Hetland et al., 2023b).

K. pneumoniae subsp. pneumoniae was identified as the main subspecies within the culture positive samples from broilers and turkeys, in both 2018 (Franklin-Alming et al., 2021) and 2020. Only three samples from broilers contained K. variicola subsp. variicola. This indicates that K. pneumoniae subsp. pneumoniae is the overall dominant KpSC subspecies in the gut of poultry. Studies on K. pneumoniae from healthy poultry are scarce, making it difficult to compare results, but our findings are in line with studies on carriage in healthy humans in Norway, where K. pneumoniae subsp. pneumoniae was also the dominant subspecies (Raffelsberger et al., 2021). Additionally, K. pneumoniae subsp. pneumoniae is the dominant subspecies reported in human blood stream- and urinary tract infections (Fostervold et al., 2022) and within the gut of healthy pigs in Norway (Kaspersen et al., 2023). A previous study hypothesized that the molecular determinants of infection are similar between humans and animals (Brisse and van Duijkeren, 2005). The observations within the current study and the above-mentioned studies seem to concur with this hypothesis. Further work with a One Health perspective and focus on cross-sectional sampling is needed to further investigate this hypothesis.

Infections with K. pneumoniae are sporadically identified in animals (Brisse and van Duijkeren, 2005; Bidewell et al., 2018). There is, however, limited knowledge on the pathogenic potential of K. pneumoniae in poultry. In the present study, the hypervirulent clone ST4710 persisted within the turkey production during the study period. Moreover, the clinical ST4710 isolate from 2018 was closely related to the other ST4710 from healthy animals in 2018, and to some extent in 2020. Healthy carriers of ST4710 were also identified from the same farm within the same time frame as the clinical ST4710 isolate was detected. This indicates that the ST4710 clone identified within the cecum of healthy turkeys has the potential to cause infections. The pathogenicity of ST4710 may be due to the presence of the p4710 plasmid, which was present within all the ST4710 isolates. Interestingly, the salmochelin locus was lost in the 2020 isolates, and may be a result of evolutionary adaptation of K. pneumoniae within the turkey gut. Of the most common siderophores in hypervirulent K. pneumoniae, aerobactin is regarded as a critical virulence factor for hypervirulent K. pneumoniae (Russo et al., 2015). In contrast, salmochelin was found to be the least important siderophore. The loss of the salmochelin locus in the 2020 ST4710 isolates may therefore be related to fitness, since salmochelin does not contribute significantly to siderophore production in the presence of the aerobactin locus (Russo et al., 2015). The presence of a yersiniabactin locus in the ST4710 isolates from both 2018 and 2020 may also support this hypothesis, as the siderophore yersiniabactin has also been found to have a higher activity compared to salmochelin (Russo et al., 2015). According to the theory of bacterial streamlining, bacteria evolve by reductive evolution, and thus try to reduce their genome size for efficient nutrient use (Giovannoni et al., 2014). The presence of both the aerobactin and yersiniabactin loci may therefore have rendered the siderophore production of salmochelin redundant and of little use to the bacterium. The loss of virulence genes in favor of the aerobactin locus has previously been described in carbapenem-resistant hypervirulent K. pneumoniae (Shankar et al., 2021), where the loss of rmpA2 was detected. The presence of both rmpA2 and the aerobactin locus within the same isolate led to slower growth (Shankar et al., 2021), increasing the fitness cost. Further studies are needed to confirm if a similar hypothesis can be applied to the salmochelin and aerobactin loci.

A comparable occurrence of AMR genes were identified in 2020 and 2018 (Franklin-Alming et al., 2021). As before, genes conferring resistance to tetracycline were the most prevalent within the poultry production, however, the overall prevalence of AMR was low in both the turkey and broiler flocks. The five major STs that persisted from 2018 to 2020 did not carry any AMR genes, except ST590, which had a similar profile as that detected in ST590 isolates in 2018. Thus, the presence of AMR genes may not be an explanatory factor for the persistence of these STs within the poultry production.

In conclusion, several STs persisted within the broiler and turkey production within the study period, the major ones being ST17, ST35, ST37, ST590, and ST4710. This persistence may be due to local re-circulation or reintroduction from parent flocks, as described in several other studies on E. coli. Most of the persisting STs did not carry AMR genes, and the persistence of these STs within the poultry production is therefore likely not associated with the presence of such genes. The detection of a clinical ST4710 isolate in 2018, highly similar to other ST4710 isolates identified in the gut of healthy turkeys within the same time frame and location, highlight the transmission- and infectious potential of ST4710 in turkeys. The loss of the salmochelin operon in the ST4710 isolates from 2020 may be a result of reductive evolution due to the presence of other siderophores within the same isolates.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary material.

Ethical review and approval was not required because the samples were taken after routine slaughter, under the auspices of the NORM-VET surveillance program.

IL, AU, and MS provided the funding and performed the project administration. MS, AU, and HK conceptualized the study and wrote the initial manuscript. FF-A, HI, EB, and HK did the laboratory analyses. HK and MH analyzed the data. All authors contributed to editing the manuscript.